Acconcia, Corsetti and Simonelli - 2014 - Mafia and Public Spending - Evidence On The Fiscal Multiplier From A Quasi-Esperiment

American Economic Association
Mafia and Public Spending: Evidence on the Fiscal Multiplier from a Quasi-Experiment
Author(s): Antonio Acconcia, Giancarlo Corsetti and Saverio Simonelli
Source: The American Economic Review, Vol. 104, No. 7 (JULY 2014), pp. 2185-2209
Published by: American Economic Association
Stable URL: https://www.jstor.org/stable/42920883
Accessed: 02-03-2020 18:51 UTC
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
American Economic Association is collaborating with JSTOR to digitize, preserve and extend
access to The American Economic Review
This content downloaded from 137.222.134.31 on Mon, 02 Mar 2020 18:51:02 UTC
All use subject to https://about.jstor.org/terms
American Economic Review 2014, 104(7): 2185-2209
http://dx.doi.org/10. 1257/aer. 104. 7.2185
Mafia and Public Spending:

Evidence on the Fiscal Multiplier from a Quasi-Experimenť
By Antonio Acconcia, Giancarlo Corsetti, and Saverio Simonelli*
A law issued to combat political corruption and Mafia infiltration of

city councils in Italy has resulted in episodes of large, unanticipated,
temporary contractions in local public spending. Using these epi-
sodes as instruments, we estimate the output multiplier of spending
cuts at provincial level - controlling for national monetary and fiscal
policy, and holding the tax burden of local residents constant - to be
1.5. Assuming that lagged spending is exogenous to current output
brings the estimate of the overall multiplier up to 1.9. These results
suggest that local spending adjustment may be quite consequential
for local activity. ( JEL D72, E62, H71, K42)
The widespread resort to fiscal stimulus at the onset of the global crisis and, more
recently, the emerging need to consolidate deficits in response to rising fiscal imbal-
ances have revitalized the empirical debate on the transmission of fiscal policy -
"the multiplier." While the literature has mostly focused on aggregate effects at a
national level, several recent contributions (reviewed below) have called attention
to the local dimension. This shift in focus is motivated by specific policy questions,
combined with the opportunity to exploit institutional information to address econo-
metric issues in identification.1
♦Acconcia: Department of Economics and Statistics, University of Naples Federico II, Via Cintia, 80126 Napoli,
Italy, and CSEF (e-mail: antonio.acconcia@unina.it); Corsetti: Faculty of Economics, Cambridge University,
Sidgwick Avenue, Cambridge, United Kingdom CB3 9DD, and CEPR (e-mail: gc422@cam.ac.uk); Simonelli:
Department of Economics and Statistics, University of Naples Federico II, Via Cintia, 80126 Napoli, Italy, and
CSEF (e-mail: saverio.simonelli@unina.it). We would like to thank three anonymous referees, our discussants
Luigi Guiso, Jayant Ganguli, Veronica Guerrieri, Tullio Jappelli, seminar participants at Banca di Italia, Bank
of Albania, Cambridge University, Chicago Booth, New Economic School (Moscow), the 2010 SIE conference,
the 2011 IMF-EUI Conference on Fiscal Policy, Stabilization, and Sustainability at the EUI, the 26th Annual
Congress of the European Economic Association (Oslo), the 43rd Annual Meeting of the Money, Macro and
Finance Research Group (Birmingham), and the third workshop of the fRDB Fellows and Affiliates, as well as
the contributors to the New York Times blog, for useful comments and discussions. We would also like to thank
the "Ministero dell'Interno - Dipartimento per gli Affari Interni e Territoriali" and Eva Belli, as well as the office
of the major, and the financial department of Pompei for data on city council dismissals. Jasmine Xiao and David
du Plessis provided superb research and editorial assistance. The work on this paper is part of PEGGED (Politics,
Economics and Global Governance: The European Dimensions), Contract no. SSH7-CT-2008-2 17559 within the
seventh Framework Programme for Research and Technological Development. Support from the Pierre Werner
Chair Programme at the European University Institute is also gratefully acknowledged. The authors declare that
they have no relevant material or financial interests that relate to the research described in this paper.
ŤGo to http://dx.doi.Org/10.1257/aer.104.7.2185 to visit the article page for additional materials and author
disclosure statement(s).
Multipliers are typically estimated by tracing the effects of exogenous fiscal impulses on economic activity.
Much of the debate has focused on identifying innovations in spending or taxation, distinct from variations that are
systematically related to the business cycle. Failure to draw a sharp distinction in this dimension means that reverse
causation from output to spending and taxes, coupled with possible anticipation effects, may spuriously raise (or
lower) estimated multipliers (see, e.g., Blanchard and Perotti 2002, and Ramey 201 1).
2185
2186 THE AMERICAN ECONOMIC REVIEW JULY 2014
A key question concerns the efficacy of fisc

recessionary shocks, which would entail a re
regions. A related question concerns the g
quences of crises that may force local admin
different intensities. The body of evidence f
no guidance on these issues. Compared to na
cial economies are much more open and face
that, being set at the national level, is large
conditions.
In this article, we provide evidence on output multiplier effects of government
purchases at a local level, relying on a quasi-experiment. Focusing on public invest-
ment in Italian provinces, we instrument spending by exploiting an Italian law
which, upon evidence of Mafia infiltration in a city council, mandates the dismissal
of all elected officials, who are replaced by three external commissioners appointed
by the central government. The instrument builds on the fact that (i) the police inves-
tigation and the emergence of the incriminating evidence leading to a city council
dismissal are unrelated to fluctuations in local economic activity; and (ii) the com-
pulsory administration by external commissioners, after the dismissal of elected
officials, typically translates into an immediate, unanticipated, and temporary cut
of public investment projects. The first year of compulsory administration, indeed,
records a strong contraction in provincial public spending, with an average drop
of 20 percentage points (corresponding to about one-half of a percentage point of
provincial value added). In addition, due to the characteristics of fiscal federalism
in Italy during our sample period, variations in public expenditure in a municipality
cause little or no variation in the tax burden faced by residents, since virtually all
local spending is financed by transfers from the central government. Hence, we are
able to estimate multipliers of local spending controlling for the aggregate business
cycle and the national monetary and fiscal policies (with fixed effects), and indepen-
dent of the implied adjustment in taxes.
In our findings, the contemporaneous output multiplier of spending contractions -
not compensated by monetary expansions, holding the tax burden constant - is as
high as 1.5. Furthermore, under the maintained assumption that lagged spending
is exogenous to current output, the combined effects of past and current spending
bring our multiplier estimate up to 1.9. We also find no significant spillovers of pro-
vincial spending into adjacent areas, suggesting that local economies may actually
be quite "insular" from each other.
A key concern in our analysis is the possibility that city council dismissals for
Mafia infiltration may affect output via channels that are independent of spending
cuts, in violation of the exclusion restriction. We specifically address two potential
channels: (i) variations in the scale of Mafia activities in response to a dismissal,
as some of these activities may directly or indirectly affect provincial value added;
and/or (ii) a "shock to government" induced by the replacement of elected officials
with external commissioners, as the change may result in a slowdown in the issu-
ance of licences to build or permits to start new businesses, or in a hiring freeze. As
regards the first channel, our regressions include controls for the variations in the
size of Mafia activities during a compulsory administration, capturing the outcome
of police investigation and legal action against mobsters (such as arrests of, and
VOL 104 NO. 7 ACCONCIA ETAL: MAFIA MULTIPLIER 2187
charges against, mobsters for Mafia-related crimes) ove

ing these controls from our regressions tends to reduc
suggesting that the Mafia-activity channel and spendin
effects on short-run economic activity. As regards th
data on the universe of city council dismissals in Italy
motivated by Mafia infiltration are not associated with
have no effects on output.
Together with the present study, a number of recent
the analysis of output multiplier effects using subnatio
relative to national-level military spending in the U
Steinsson (2014) estimate multipliers in the range 1.4-
Serrato and Wingender (201 1) use fund reallocation ac
sions in the estimates of local populations following ch
odology, while Shoag (2010) exploits the idiosyncratic
on defined-benefit pension plans managed by the US
multipliers are as high as 1.88 and 2.12, in the respec
Fishback and Kachanovskaya (2010) exploit a swing vo
primarily across US states, to instrument government
In their results, the point estimate of the multiplier fo
Similar to these studies, in our empirical model we co
policy and wealth effects from tax adjustment. Relativ
in addition to using non-US data, our contribution ha
tures. First, our analysis disentangles impact and dyna
Second, although our regression model does not expli
effects of spending increases and cuts, our estimates o
on sharp fiscal contractions. Finally, we should stress
consists of government purchases, instead of transfers
The rest of the article is organized as follows. Sectio
model. Section II is devoted to the analysis of our ins
institutional details on the laws targeting Mafia conn
our main results. Section IV provides evidence that ou
is not contaminated by transmission channels unrelat
Section V discusses results for alternative specificatio
Section VI concludes.
I. The Empirical Model
In our study, we aim to recover the short-run multiplicative effects of public

spending on output at the provincial level in Italy. We present the regression model
in this section and discuss our instrument in the following section.
2 An output multiplier of about two is also implied by the estimates of Chodorow-Reich et al. (2012), looking
at the employment effects of state fiscal relief. Analogously, large employment local effects are found by Moretti
(2010). Nonetheless, multipliers are found to be not significantly different from zero by Clemens and Miran (2012),
who build on differences in the balanced-budget requirements at the state level. Cohen, Covai, and Malloy (201 1),
who instrument public spending with changes in congressional committee chairmanship, note that spending varia-
tions appear to significantly dampen corporate sector investment and employment activity. However, as suggested
by the authors themselves, their results may reflect the high level of employment prevailing in their sample.
To carry out this study, we have assembled a d

ment spending in each province of Italy over t
1999, a period over which we could obtain comp
on public works. An Italian province is a geogra
and contains several municipalities. During this
Italy; hence, we have 950 annual observations.
For each province, let y¡ denote the real per c
of growth, defined as Yit = (y¡,t - y¡,t-i)/y¡,t-ú
capita public investment in infrastructure, and G
of lagged value added, Gi t = (git, - g¡t,-{)/y¡tt-i
(see, e.g., Barro and Redlick 2011), we estimate
the growth of per capita value added in a provin
in per capita spending on infrastructure in the
model is
(1) Y¡¡ = ßGu, + a¡ + A, + 7 X, , + v,-,,
where the coefficient ß measures the contemporaneous one-year government spend

ing multiplier; a, is a province fixed effect; A, is a year fixed effect; and X denotes
vector of further control variables, to be discussed below.
The inclusion of a year fixed effect in equation (1) serves two main purposes.
First, it controls for national components of public investment and GDP common
to all provinces. As variations in aggregate spending and output are usually predict
able and arguably endogenous to cyclical developments, they may lead to spuriou
estimates of the multiplier due to reverse causation.
Second, year fixed effects control for monetary and fiscal policy at the national
level. As is well understood, the transmission of fiscal stimulus or contraction is
bound to be crucially affected by the monetary stance, as well as by the anticipatio
of fiscal measures (spending cuts or tax hikes) dictated by the need to stabilize pub
lic debt in the medium and long term (see, e.g., Christiano, Eichenbaum, and Rebel
2011; Corsetti, Meier, and Muller 2012b; and Woodford 2011). Failure to control
for these factors means that the estimated multipliers conflate the effects of fisc
shocks with those of the monetary and budget policy that is anticipated to prevail
over both short- and long-term horizons.3
Through province fixed effects we address potential endogeneity issues raised
by the possibility that province-specific characteristics may be correlated with
spending allocation criteria. By way of example, it may be possible that the centra
government systematically allocates relatively large projects in lower-growth prov
inces in an effort to spur local economic activity. Under this allocation criterion, th
OLS estimates of the multiplier would tend to be spuriously low.
An advantage specific to our data relates to the type of fiscal federalism in Italy
during our sample years, based on Law No. 281/1970 and Law No. 382/1975. O
3 The challenge of estimating aggregate multipliers while accounting more explicitly for budget and monetary
policy has been taken on by a new generation of contributions (see Corsetti, Meier, and Muller 2012a; Canova and
Pappa 201 1; Ilzetzki, Mendoza, and Vegh 2013; Leeper, Walker, and Yang 2009, among others). Heterogeneity
consumption responses to fiscal stimulus is explored by Misra and Surico (2013).
VOL. 104 NO. 7 ACCONCIA ETAL: MAFIA MULTIPLIER 2189
the spending side, these laws gave the central governme

overall flow of resources accruing to local governments
full control of these funds, including the power to sel
firms to carry them out. On the revenue side, however,
little power to set tax rates.4 Therefore, throughout o
resources channeled by the central government into loc
not matched by variations in the tax burden of the loca
we do not face potential issues arising from the omiss
from our set of controls.6
As regards the matrix of controls, X, we include five
number of people reported to the judicial authority fo
(ii) extortion; (iii) Mafia-related murders; (iv) corruptio
corruption crimes reported to the judicial authority.7 A
in per capita difference terms and entered in the regres
neously and lagged up to two years. As argued below, to
council dismissals coincide with intense police investig
scale of Mafia activities, crime deterrence implied by a
economic activity in a province independently of cuts
variables defined above are included in order to control f
sion channel, under the maintained assumption that th
correlated with the outcome of police investigation in te
ber of people charged with Mafia-related crimes. We sh
cross-sectional perspective, areas where the Mafia pres
likely to be characterized by a relatively high average n
by the police. However, the inclusion of province fixed
cross-province differences in these averages. By the sa
the degree of law enforcement may vary over time, du
political or media pressure, changing priorities of the law e
judges and prosecutors. In our analysis year fixed effec
in enforcement over time.
To control for local cyclical conditions, we include lagged changes of two prox-
ies for unemployment, namely, the (t - 1 and t - 2 log difference of the) per capita
in our sampic years, ine iianan municipalities nau uie upuuii lu iiiaigmaiiy aujusi uic laic ui iwu laAcs sci ai
national level, to address local financial needs. The revenue from these adjustments (if any) nonetheless accou
for a very small share of their overall budget.
f..n,4;nn tWo rrA„0rnm0nf ÍCn oeoeö
nui auipuaiugijr, luwai guvçiiiuiçiud iuuuiçu lui puuiiv luuuiug iiuiii tiiv vvuuai guvviiiuivui
1983). Their success was helped by the fact that, historically, public investment w
instrument to foster growth and sustain social cohesion.
0'17UI1/i fUa *v' /-vť fVi â fnv *v' ii1fî«>1î ¿vrr nfir'rt /m i frti i f +s*' tv* nrrti ol fov rnfor r'ir fov rmrannao « o
ty iiiiç mv iiiaguuuuç 'j i luv laA uiuiupiiud
controversial, recent empiri

output - see, for instance, B
sizes that aggregate spendin
political concerns with the o
on output, theses authors ar
spending multiplier.
v^uuupuuii Clinics liiwiuuc ciiiuc¿¿iciiiciu, uiiàappiupiiauuu ui puuui luiius, caiuwuii, miu uiiuciy agiccuicuu».
The categories (i) through (v) are used in the reporting of official statistics by the "Istituto Nazio
ISTAT, according to the classification of crimes in Italian law. The first three categories strictly refer
(see "Codice di Procedura Penale art. 51, comma 3 bis."), while the last two are related to corr
officials. In particular, article 416-bis of the Italian penal code defines the crime of Mafia-type a
murders related to Mafia activity are recorded by ISTAT according to information supplied by the
employment, and hours of wage supplement

ance scheme available to employees of large pr
Guadagni).8 Including these controls is especi
are highly persistent (Chodorow-Reich et al.
the t - 2 and t - 3 lags of the number of m
istration in the province, weighted by the re
our spending variable Gi t.
The key identifying assumption in the SVAR
(2002) is that lags of Gi t are predetermined
assumption, the coefficients on the lags of G
of the dynamic multiplier, complementing ou
multiplier.
The provinces in our panel have different si
our regressions are weighted by province-le
known, inference in panel estimation can be
correlation within groups of observations, o
Bertrand, Duflo, and Mullainathan 2004; Ang
serial correlation problem, we will use up to
Regarding the spatial correlation problem, f
(2004), we posit that provinces belonging to
result of an unobserved cluster effect due t
Our inference will therefore be based on sta
ous spatial correlation allowing for 190 clust
19 regions: because of its small size, we aggr
Piedmont), as well as robust to heteroskedast
II. Instrumenting Changes in Public S
Despite the advantages of our empirical m

tion of equation (1) would expose our results
infrastructure is usually planned some years
account for anticipation effects over the tim
the realization of projects can substantially b
Second, in our sample, the government may
local developments, in ways that are not acc
To address these problems, we need a good in
public spending.
8 The Cassa Integrazione Guadagni (CIG) is an Italian institut

supporting large firms in a temporary crisis. It provides temp
lost their jobs or are forced to work for reduced hours.
Dropping these variables from our preferred specification, h
10 Apart from minor numerical differences in the point estima
if we do not weight our regressions (see Acconcia, Corsetti, an
VOL 104 NO. 7 ACCONCIA ET AL: MAFIA MULTIPLIER 2191
A. The Institutional Setting: Mafia Infiltration and Co
We introduce our instrument by providing backgrou

the Italian law deals with Mafia-related crimes. In view
organized crime in the Italian economy, two articles we
1982, explicitly targeting Mafia-type organizations." A
target the use of intimidation, associative ties, and
acquire direct or indirect control over otherwise legal e
in relation to public investment and the provision of
mentioned, the distorted incentives created by the laws
the 1970s and the end of 1990s favored a strong growth
During our sample years, indeed, public works managed
Italy became one of the most lucrative sources of busin
The rise in Mafia infiltration of public administration
arguably a key motivation for introducing tougher anti
1990s. Among these measures, a law was passed allowin
to remove elected local officials on evidence that their decisions were determined or
influenced by the Mafias (D.L. 3 1 /05/ 1 99 1 n. 1 64) . According to this law, upon the
removal of a city council, the central government appoints three nonelected, exter-
nal commissioners, who govern the municipality for a period of 18 months.
This new law gave prosecutors a key new tool to combat the Mafia, sharply increas-
ing the value of police investigation. Before its introduction in 1991, incriminating
evidence against, say, the alderman of a city, would lead to the arrest of an individual.
After 1991, the same evidence could lead to the dismissal of the entire city council,
thus creating opportunities to fight the networks connecting Mafia-controlled firms
and public administrations.
Not surprisingly, the new tool has been extensively (although not exclusively)
used in regions where criminal infiltration in the territory and the institutions is
long-established and common knowledge. As shown in Table 1, over the years
dismissals have been mostly concentrated in the provinces of Naples, Palermo,
Reggio Calabria, and Caserta. The geographical distribution of the Mafia varies
both across and within regions. It is highly concentrated in the southern regions of
Sicily, Campania, Calabria, and Puglia, but is also significant in northern regions
like Piedmont and Lombardia. Within these regions, in turn, there are substantial
1 ' Historically, different Mafia-like organizations have been active in different Italian regions: the Camorra in
Campania, the 'Ndrangheta in Calabria, the Sacra Corona Unita in Puglia, and the Mafia in Sicily. Each organiza-
tion in turn comprises different groups or clans, with the best known being the Cosa Nostra in Sicily and, recently,
the Casalesi in Campania.
12 See Acconcia et al. (forthcoming), and references within, on the influence of Mafias on the legal economy via
their relations with public officials, including political representatives, judges, local administrators, and members
of the police force.
13 An important role was also played by the strong earthquake that hit the south of Italy at the end of 1980. With
the need to reconstruct housing and infrastructure, a large inflow of government funds benefited areas of Italy tra-
ditionally under the control of the Mafias.
14 According to official estimates (Ministro dell'Interno 2000), over our sample period the profits accrued to
organized crime from controlling public works were comparable to those of extortion and drug dealing. Mafia
infiltration has created a vast network connecting legal and illegal activities. For instance, the Commissione
Parlamentare di Inchiesta (2005) emphasizes that many firms suspected of Mafia collusion operate with high stan-
dards of efficiency across the country.
Table 1 - Council Dismissals because of M ah a Infiltration
Napoli 48 Reggio C. 37 Palermo 23 Bari 5

Caserta 31 Catanzaro 8 Catania 9 Lecce 2
Salerno 6 ViboV. 12 Trapani 6
Avellino 4 Crotone 3 Caltanisetta 6
Benevento 1 Cosenza 2 Agrigento 7
Messina 3
Ragusa 1
Campania 90 Calabria 62 Sicily 55 Puglia 7
Notes: The table reports the number of council dismissals because of M

1991-2012 (July), by province, within the regions of Calabria, Campa
Only seven council dismissals occurred in the rest of Italy during the sa
differences across provinces, mostly driven by historic

own strategies and pervasiveness (see, for instance, Dick
Our sample includes 110 cases of city councils put unde
tion for Mafia infiltration, but since we carry out our s
aggregating these cases by province, we obtain 47 obser
B. An Instrument "One Can't Refuse"
When the government of a municipality is dismissed o

tration, the external commissioners appointed by the ce
cut financial flows into local public works and investm
indeed, the first year of compulsory administration in
with a sharp contraction in spending on public works a
shown in Table 2, in which we compare changes in pub
with and without municipalities under ongoing compulso
As the treatment group, we pool together all the provi
of ongoing dismissal and compute the change in investm
lowing the publication of the dismissal decree. As contro
all the province-year observations not in the treatment g
sample (columns 1 and 2 of the table), or the subsample
one dismissal (columns 3 and 4). The rationale for defin
groups is to show that the mean differences are not driv
neity in average spending changes across provinces whi
cases of compulsory administration.
Columns 1 through 4 in Table 2 show that the mean d
between the treatment and each control group is negativ
cant at the 5 percent significance level. The average cont
15 Statistics on convictions of the crime of Mafia association by regions an

geographical differences in the presence of the Mafias. Namely, 90 percent o
2001 were put on trial by courts in the Southern regions - mainly Sicily,
were, however, significant differences within each region. In the Campania r
victed in the judicial district of Salerno (corresponding to 24 convictions per
the district of Naples (32 convictions per 100,000 inhabitants). In the Calabri
Reggio amounted to, respectively, 204 and 343 (that is, 14 and 59 per 100,000
corresponding numbers in Bari and Lecce were 142 and 534 (6 and 30 per 100
convictions were handed down by courts in Piedmont and Lombardia.
VOL 104 NO. 7 ACCONCIA ETAL: MAFIA MULTIPUER 2193
Table 2 - Investment Spending in the First Year after Council Dismissal
Difference -19.65*** -0.46** -23.67*** -0.49* -4.72 -0.04

[5.36] [0.19] [7.12] [0.26] [5.29] [0.18]
Observations 950 950 180 180 905 905
Notes: The table reports one-sided mean difference

the treatment and control groups, columns 1-4, as
trol groups, columns 5 and 6. Investment changes
ment, columns 1, 3, and 5, or lagged value added,
consists of province-year observations in the first
The control group consists of the rest of the sampl
which never experienced local government dismiss
to 1999 at provincial level. The standard errors are
*** Significant at the 1 percent level.
** Significant at the 5 percent level.
♦Significant at the 10 percent level.
treatment group amounts to about half

comparable the change in fiscin size to
multipliers.16 two columns in t The last
ing variations are not statistically differ
with the assumption that treatment and
their treatment status.
To gain insight into how the dismissal of
spending in practice, we collected exten
(within the province of Naples).17 The c
2001, following the arrests of the speak
alderman for street maintenance for Mafia association. The councilman was identi-
fied as the main link between the local administration and the boss of the Mafia clan
operating in Pompei, who was also arrested in the course of the same investigation.
The extent and type of spending cuts associated with a dismissal are best ana-
lyzed via a detailed comparison between the (ex ante) annual budget, and the actual
expenditure flows. In the case of Pompei, the 2001 budget prepared by the elected
officials before the dismissal had allocated 4 million euros to public works. Upon
taking over the city administration, the commissioners formally ratified the budget
but, at the same time, cut spending on public works by more than 3 million euros.18
During 2001, actual spending amounted to a mere 20 percent of that planned.
The spending cuts affected a variety of budget chapters: (i) extraordinary street
maintenance; (ii) improvement and maintenance of the public lighting system;
16 As regards defense spending, changes in fiscal variables related to the Korean War amounted to 5.6 percentage
points of GDP in 1951, 3.3 in 1952, and 0.5 in 1953, followed by a contraction of 2.1 percentage points of GDP in
1954. Changes that occurred during the Vietnam War amounted to - 1.2 and 1.1 percentage points of GDP in 1966
and 1967, respectively (see Barro and Redlick 201 1). On the revenue side, the effect of the 54 legislated exogenous
tax changes identified by Romer and Romer (2010) amounts to -0.03 percentage points of GDP.
17 While the Pompei dismissal was just after the end of our sample, it was the case for which we were able to
obtain the richest and most accurate information from a variety of sources, including interviews with local admin-
istrators and commissioners.
18 From an accounting point of view, this was accomplished by moving three million euros of investment to the
item economie , that is, savings on expenditures.
(iii) purchase of mechanical equipment; (i

maintenance of the water system; (vi) maint
(vii) extraordinary maintenance of the sewa
(ix) municipal cemeteries. Not surprisingly,
investigation, or under the control of the cit
ation. However, the commissioners also decid
arguably with the objective of acquiring more
spending decisions.
C. Is the Instrument Variation Systematically
Police investigations leading to the dismissa

infiltration may be conducted for a variety o
trol of local public works. Based on the repo
d'Inchiesta to the Italian parliament, dismiss
gations of crimes by local administrators or
their official functions); (ii) investigations of
drugs, and Mafia wars for the control of loca
by whistleblowers, providing information o
infiltration in public administration; (iv) inve
of a city mayor or a city council member, su
Parlamentare d'Inchiesta 2005). The same doc
dismissals are not prompted by indicators of
curement procedure. On the contrary, it is of
cedures involving firms connected to the Ma
price, with no apparent waste of public resou
The account of the circumstances leading
Commissione Parlamentare d'Inchiesta sugg
missals and local economic activity, as the in
randomly in the course of ongoing police in
formal statistical evidence, we compare the gr
to their first dismissal, with the growth rate
dismissal, by running the following regressio
(2) Y¡ f = do + d i D¡¡ + dļt + dļ(t x D¡¡ ) +
where fis a time trend and D¡ , is a dummy variable equal to 1 for any province x year
observation before the first episode of council dismissal and 0 otherwise. Based
on the sample 1986-1999, OLS results yield a point estimate of the coefficient d3
19 By way of example, city council dismissals followed the arrest of local administrators on charges of drug
trafficking in Roghudi (province of Reggio Calabria), and Cesa (province of Caserta); and the arrest of the mayor
and members of the city council on charges of theft, infringement of building laws, and bid rigging in Sant'Andrea
Apostolo dello Ionio (province of Catanzaro). In a few cases (e.g., Gioia Tauro, province of Reggio Calabria), the
mayor was explicitly charged with the crime of Mafia association. City council dismissals followed from investiga-
tions of deadly Mafia ambushes in Lametia Terme and Guardavalle (province of Catanzaro), and the investigation
of threats against local administrators in Bordighera (Imperia). Direct and indirect links between local administra-
tors and criminal organizations were uncovered after the resignation of city council members and/or the mayors
of Taurianova and San Ferdinando (Reggio Calabria), Sant'Onofrio (Vibo Valentia), and Frattamaggiore (Napoli).
VOL. 104 NO. 7 ACCONCIA ETAL.: MAFIA MULTIPLIER 2195
Table 3 - Provincial Growth Rates Prior to Council Dismissals
Above average Below average Switching
t - 1 and t - 2 1/3 1/6 1/2

t - 1 and í - 2 and í - 3 1/9 0 8/9
Notes: For any province that experienced cases of compulsory

GDP growth rates in the two and three years before the first dis
proportion of provinces for which these growth rates were always
tuated around the national average.
equal to -0.07, with a standard error adjusted for c

null hypothesis d3 = 0 cannot be rejected (the p- val
absence of a differential trend in growth rates bef
also note that we cannot reject the null hypothesis
age growth rate of the treated provinces is no differen
In a related exercise, we verify whether the grow
added in the years preceding a council dismissal is
the national average. Results are shown in Table 3,
one council dismissal.20 As apparent from the table
from the data.
A key feature of our instrument is that the time s

dence of Mafia infiltration to the replacement of the c
missioners is quite short - in our sample, it is often
takes two months.21 Hence, conditional on the news
been set in motion, anticipations of government-m
are unlikely to play a significant role in our sample
D. Implementation
We implement our IV strategy accounting for two

gate information at provincial level, the effects of
municipalities in the province need to be appropria
missal of a city council can occur at different times
investment spending, and in turn its possible effec
local value added, may crucially depend on how clo
end of the calendar year.
We use two instruments. The first instrument, d
(CDS1), equals the number of municipalities put un
provided that the official decree is published in th
weighted by the share of the province population
Including the number of municipalities proxies for
20For provinces with repeated cases of dismissals, we consider only th

from possible lagged effects of spending cuts implemented during pre
According to the law, the dismissal of a city council should normally
of the Republic. However, there are circumstances under which the loca
tative of the central government in the territory) can process the dismissa
mal decree. This speedy procedure has indeed been common practice in
2196 THE AMERICAN ECONOMIC REVIEW JULY2014
finance may be cut by the commissioners. Rel

the relative economic importance of the area un
second instrument, "Council-dismissal-S2" (
that, for each case of compulsory administratio
days between the dismissal of the city council
over all municipalities in the same province x y
observation for which this average is less than
the number of municipalities under compulsory
In particular, we instrument Gi t with "Counc
lag of "Council-dismissal-S2." Thus, the first-st
fication is
(3) G¡ , = Õ ļ CDS 1 , i + ¿2 CDS2i t_' + o
III. Results: Impact and Dynamic Mult
The results from our baseline model are shown

as well as the 2SLS estimates for two version
other including lagged values of the endogenou
The first two columns of the table show the O
on the contemporaneous and one-year lagged p
significant, but small in magnitude. In particula
0.2. As already mentioned, it is well understoo
plier of spending on public projects are potent
bias. On the one hand, there are usually long la
fund allocation and the implementation of local
takes place over several years. Empirical models
effects, which may bring forward in time var
other hand, public projects may be systematica
expansion or contraction. As OLS estimates con
policy to income with the fiscal multiplier, the
(downward) bias is more likely and larger w
cyclical.
The results from our IV model are shown in the last four columns of Table 4.
The first-stage regressions in columns 3 and 5 confirm the point suggested by
the evidence in Table 2: on average, provinces under compulsory administration
experience a sharp drop in public investment.23 As is apparent from Table 4, the
estimates of the coefficients on both instruments are negative, as expected, and
highly statistically significant. The value of the F-statistic testing the power of
the instruments is 12.58 and 1 1.83 for the model without and with lagged output
growth, respectively.
In the second-stage regression, the coefficient on contemporaneous spending
is statistically different from zero at the 1 percent significance level, with a point
22 Results are robust to instrumenting public spending without weighting the number of dismissals using the
relative population size of the municipalities under compulsory administrations.
23 We compute the 2SLS estimators using variables in deviation from province and year averages.
Table 4 - Public Spending Multplier
OLS 2SLS 2SLS
First Second First Second

stage stage stage stage
G(t) 0.21** 0.23** 1.46** 1.55***

[0.07] [0.07] [0.49] [0.43]
G(t - 1) 0.22** 0.26** -0.41*** 0.73*** -0.41*** 0.79***
[0.08] [0.08] [0.07] [0.21] [0.07] [0.19]
G{t- 2) 0.00 0.04 -0.13* 0.14 -0.13* 0.19
[0.07] [0.07] [0.06] [0.11] [0.06] [0.11]
Y(t- 1) -0.16* 0.03 -0.20**
[0.06] [0.02] [0.06]
Y{t- 2) -0.03 -0.02 -0.02
[0.05] [0.02] [0.05]
CDS'(t) -2.07*** -1.97***
[0.54] [0.56]
CDS2(t - 1) -4.02*** -4.08***
[0.98] [0.94]
F-stat instruments 12.58 11.83
Observations 950 950 950 950 950 950
Notes: Data are annual from 1990 to 1999 at the

the OLS and the second-stage 2SLS regression is t
value added divided by the previous year's per c
year-on-year change in per capita real infrastructur
the national GDP deflator) divided by the previous
and G(t - 2) are the lagged values of G. All estim
first two lags of employment and the hours of "ca
ita log-difference), the number of municipalities p
given province at t - 2 and / - 3, and a set of fiv
people reported to the judicial authority because of
Mafia murders; (iv) corruption; and (v) the number
cial authority (all specified in first-difference, in p
is by two-stage least-squares using Council-dismi
instruments. The standard errors clustered at the r
dasticity are reported in square brackets. Because o
denoting significance is as follows:
*** Significant at the 0.1 percent level.
* Significant at the 5 percent level.
estimate of 1.46 or 1.55, depending on w

among the controls (compare columns 4
this coefficient remains quite stable acr
enous cut in local public infrastructure
a contemporaneous reduction in local out
Relative to the OLS estimates, our IV e
of spending is about seven times larger
estimators. In our sample, a downwa
sonably attributed to the long and com
allocation and implementation of publ
spending is unlikely to be related to cy
of the effects materialize before projec
tive difference between OLS and IV e
Wingender (2011) and Nakamura and Stein

US data. The differences between OLS and IV
larger than in our study.24
An assessment of the dynamic effects of s
model including all controls, shown in the l
lag of output growth is significantly differen
multiplier including lagged output effects is
ß and one minus the coefficient of Y(t - 1)).
local public infrastructure by 1 percent of lo
reduction in local value added of 1.29 percent
Further dynamic multiplier effects can be d
tion that lagged spending is exogenous to cur
add up the coefficients on the contemporan
lagged value, which is significantly differen
impact of the first lag of the dependent vari
multiplier is as high as 1.95. In this model spe
null hypothesis ß < 1 in favor of ß > 1 at the
In closing this section, two remarks are in o
cal policy may differ across provinces, refle
this were the case, and the probability of tre
characteristics, IV regressions would deliver
spending for the treated areas, rather than
in quasi-experiments. In studies focused on p
expect the Mafia to affect the productivity of l
provinces, as Mafia involvement may caus
could also "grease the wheels" of public inves
ity, however, seems more relevant to assessin
public capital than to estimating the short-ru
Second, relative to a multiple-equation f
single-equation models like ours is that the e
ing do not take into account the possible fee
Thus, strictly speaking, our results cannot b
models - a point stressed by Sims (2010). How
investment does not react to value added chan
sion the coefficients of the two value added l
zero (see Table 4). In view of these results, t
a concern in our study.
24 As pointed out by one of our referees, potential differen

and the United States could reflect the different degrees of p
fiscal policy is arguably less countercyclical in Italy than in t
level. As already mentioned, insofar as the estimated multipli
where fiscal policy is more countercyclical.
25 In both specifications, the Anderson-Rubin test rejects th
a p- value about 0.01). Moreover, the Hansen /-Statistic is char
instruments are uncorrelated with the error term.
VOL 104 NO. 7 ACCONCIA ETAL.: MAFIA MULTIPLIER 2199
IV. Do City Council Dismissals Affect Output

Independently of Variation in Public Spending?
For the proposed IV estimation to be reliable, our instr

a clear effect on G, „ it must also be uncorrected with
on controls, i.e., it must satisfy the exclusion restriction
relating changes in value added to public spending, we sh
dismissal of the city council matters for provincial outpu
that it causes a (temporary but sharp) reduction in publi
We observe upfront that the inclusion of the province
many plausible reasons why the exclusion restriction cou
negative relationship between our instrument and the av
provincial level. By way of example, provinces in wh
when the city council is dismissed may have a below-av
of the Mafia.26 Conversely, the incidence of Mafia activ
in slow-growing provinces, given the lack of opportun
Finally, the risk of detection may be correlated with the in
Nonetheless, there are two other potential channels tha
restriction to fail despite the use of province fixed effec
discussion. The first, already mentioned in Section II, cou
impact on provincial value added of a contraction in mob
junction with a council dismissal. The second channel co
in the output of the local bureaucracy during periods of
A. Variation in Mafia Activities
The exclusion restriction might not hold if the empir

for the possibility that variations in Mafia activities du
dismissal affect output above and beyond variations in p
cal grounds, a successful war on the Mafia can be expec
activity in the long run, mainly through a rise in inves
entrepreneurs. Since a key objective of city council dis
reduce the presence of the Mafia in the country, one cou
trol for this channel may produce a downward bias in th
the negative output effect of a contraction in public sp
part compensated by the positive effect of "less Mafia."
For our purpose, the root of the concern with a Mafia-a
different forms of legal action and police work weighs o
of the Mafia in the short run, affecting contemporane
of strong legal action during the compulsory administrat
be signed unambiguously. On the one hand, the remova
nected to the Mafia, and a reduction in political corrupti
tions, which act like a "tax" on firms and households, ma
26Country-level studies suggest a negative relationship between the diffusi

nomic growth. Via corruption, the historical presence of the Mafia in a provin
long-run growth, which would be reflected in low output growth rates over o
the short run. On the other hand, the Mafia ma

that were previously generating value added, th
any relocation of Mafia activities across province
the level of aggregation we use in the analysis.
and difficult issues in identification. On empiri
controls that can successfully account for variat
As discussed in Section I, in our estimation
Mafia-activity channel using measures of the out
level. The idea underlying this strategy is that
ties - some of which may in principle contribute
value added - the arrest of mafiosi and intense p
changes in either or both types.27 In this section, w
ing whether our estimates of the spending multi
controls from the regression model. The exerci
direction of the net effect on output of a Mafia
trols for police investigation are omitted, the es
suggest that, overall, changes in Mafia activitie
could translate into output losses. In this case, om
misattributing the contraction of economic activ
The first column of Table 5, however, shows
multiplier falls when we exclude the controls for
output effect of a spending contraction become
effect of the Mafia-activity channel appears to p
relative to cuts in public spending. We stress t
are correlated with the instruments, as can be sh
regression of our model:
(4) Yit = <5, CDS'it + ô2CDS2it_x + a¡ + A, + çXu +
When X includes all the controls, as in our baseline specification, the effect of
the instruments on output change is statistically significant and negative. The
F-statistic testing the joint hypothesis that ¿1 = S2 = 0 is 5.13 (p- value = 0.0067).
When the controls for Mafia activity are omitted, the point estimates of ^ and S2
become less negative, i.e., they increase towards zero. Their statistical significance
is affected, too. The F-statistic testing the hypothesis ^ = 52 = 0 drops to 2.77
(p- value = 0.0651). 28
These results provide evidence that, as expected, our controls for the Mafia-activity
channel are correlated with the instruments. They also provide evidence that the
overall short-run output effect of police investigation against the Mafia, if any, is
positive - the same sign as it is commonly argued in studies of its long-run effect.
27 For our measures to be a good proxy for the Mafia-activity channel, the scale of Mafia activities must be cor-
related with arrests and charges of mafiosi by the police. Arrests and charges may actually fall during a compulsory
administration, if overall deterrence causes mafiosi to withdraw from the province. Moreover, a fall in the number
of arrests may reflect the decision by the Mafia to downscale activities. We thank two referees for stressing these
points.
28 As a complementary exercise, we replace the left-hand side of the reduced form with Y - ßG, imposing
¡3 = 1.55 as suggested by our estimate. In this case, the F-statistic becomes very close to zero (p- value = 0.99),
consistent with the presumption that the orthogonality conditions are valid.
Table 5 - Do City Council Dismissals Affect Output

Independently of Variation in Public Spending?
G{t) 1.30** 1.56*** 1.53*** 1.59*** 1.64***

[0.45] [0.41] [0.44] [0.47] [0.44]
G(t- 1) 0.69*** 0.79*** 0.78*** 0.81*** 0.82***
[0.19] [0.19] [0.19] [0.20] [0.20]
G(t - 2) 0.16 0.19 0.19 0.20 0.20
[0.09] [0.11] [0.11] [0.11] [0.11]
Y(t- 1) -0.19** -0.20** -0.20** -0.20** -0.20**
[0.06] [0.06] [0.06] [0.06] [0.06]
Y(t - 2) -0.02 -0.01 -0.02 -0.02 -0.01
[0.05] [0.05] [0.05] [0.05] [0.05]
Resignation(r) -0.00
[0.05]
Resignation^ - 1) 0.04
[0.06]
Election(ř) -0.03
[0.08]
Election(ř - 1) 0.03
[0.13]
Budget - no confidence vote(ř) 0.03
[0.15]
Budget - no confidence -0. 1 6
vote(/ - 1) [0.16]
Others(ř) 0.45
[0.39]
Others(í - 1) 0.29
[0.39]
F-stat instruments 12.56 14.17 12.30 12.12 14.75
Observations 950 950 950 950 950
Notes: The table shows results from either dropp

adding controls capturing council dismissals for
umns 2-5). Dismissals may occur because of res
of ineligibility of the mayor; failure to organize
and political crisis in the ruling coalitions. All th
dled together in the last two regressors, account
infiltration. The standard errors clustered at the re
dasticity are reported in square brackets. Because
denoting significance is as follows:
Arguably, these results do alleviate co

not appropriately controlled for - wo
local multipliers.
B. A Shock to Government
Independently of their effects on public spending, it might be possible that city

council dismissals per se are negative shocks to the productivity of local admin-
istration. Specifically, one concern is that the sudden replacement of elected offi-
cials with external commissioners may reduce administrative output, with negative
effects on economic activity. By way of exa

issued in a municipality may drop during com
This concern turns out to be unfounded on bot
On institutional grounds, city council dismis
tive in the fight against the Mafia. The comm
as efficiently as possible, with the specific g
benefits of freeing local institutions from th
tive quotes from official documents. "The com
be an opportunity for improving the admini
between the government and the citizens" (D
administration must not be a simple bridge
tunity for the development and growth of
tunity for a new beginning for the local com
d'Inchiesta 2005, p. 9).29
Based on an in-depth analysis of a sample o
covered by our study), the report by the Min
the commissioners pursued their mandate sc
findings of the report, the external commission
(such as new hiring), which were de facto bl
tions attributable to the Mafia, were taken t
education, police force, and social work.30
To shed light on the issue, we extend our em
city councils can also be dismissed for reasons
necessarily implying a spending cut on public
shocks to government, they should have a ne
do not entail any contraction in spending.
For our model extension, we built a dataset
cil dismissals in Italy not related to the 1991
als include: (i) resignation by elected offic
(iii) special cases of ineligibility of the mayor
and (v) political crisis in the ruling coalitions
and (v) were the least common cases, we mer
We also aggregated the municipality-level inf
our main dataset.
The key result from our extended model is that the cases of council dismiss-
als not related to Mafia infiltration are uncorrelated with a drop in public spend-
ing: in the first-stage regression for our augmented model (not shown), none of the
new covariates is statistically significant at the 5 percent level. The question is then
29 Our own translation.

30 However, the official documents also recognize the limits of the commissioners' achievements, pointing out
that in most municipalities there were few or no fundamental changes by the end of compulsory administrations.
In other words, in the assessment by the Commissione , the main shortcoming of the law was not a reduction in
the output by the local administration, but the absence of any significant improvement in its performance: the
achievements by the commissioners were limited to ongoing administrative activities (Commissione Parlamentare
d'Inchiesta 2005).
During our sample period, the total number of city councils dismissed for reasons not related to Mafia infiltra-
tion was 2,031. The most common reason was the resignation by elected officials, which accounted for about half
of all cases.
whether dismissals of city councils for reasons other th

not associated with a significant variation in public spe
effect on output. As shown in columns 2 through 5 of
the estimated coefficients are not significantly different f
Summing up, neither administrative documents nor s
evidence of a "compulsory administration channel" whic
through a drop in the performance of local bureaucracy
to materialize only when dismissals are associated with
V. Further Results
In this section, we further investigate the properties of our empirical model.

Specifically, we analyze the cross-border effects of local spending on the output of
neighboring provinces, the influence of individual provinces on our estimates, and
the implications of restricting our sample to the southern regions only.
A. Cross-Border Effects
Spending variations in one province may affect economic activity in neighboring

provinces, through different channels. On the one hand, some of the contraction
in demand in one municipality may "leak" into nearby areas, driving down eco-
nomic activity simultaneously within and outside the province where spending is
cut. Demand spillovers would induce a positive correlation in the response of value
added in adjacent provinces. On the other hand, in response to a localized spending
shock, it is possible that production factors relocate, moving across the borders of
the province hit by the fiscal contraction. With this second type of spillovers, the fall
in local economic activity in the province under compulsory administration would
correspond to an increase in economic activity in the nearby areas, inducing a nega-
tive correlation in the response of provincial value added across borders. If either
type of spillover were to be empirically relevant, our estimates would miss part of
the output effects of spending innovation in a province, reflecting either demand
leakages or relocation effects.
We carry out an analysis of the cross-border effects of local spending both by
extending the set of regressors, and by aggregating small adjacent provinces. Results
are shown in Table 6. Specifically, in the first column, the regression model also
includes the variable SG¡ , = ^g'sv. ^1'' 1 and its first lag, where for province i and
year t, Sgi,t is the per capita investment in provinces which are in the same region
excluding province i itself, and the variable is defined accordingly. The coef-
ficients of the newly defined variable and its lag are low; that of SGi t is not signifi-
cantly different from zero.
In the second column of Table 6, we enter SG, interacted with G, both
measured in terms of deviation from the respective mean value. We thus allow for
the possibility that the effect of local spending reflects either complementarity (as
a result of demand leakages), or substitutability (as a result of high spatial mobil-
ity of factors of production) between spending in adjacent areas. The coefficient on
the interaction term is not significantly different from zero. Note that including the
Table 6 - Spillovers
G(t) 1.44**
1.50*** 1.24**
[0.47] [0.41] [0.45]
G(t - 1) 0.73*** 0.76*** 0.74**
[0.20] [0.17] [0.23]
G(t- 2) 0.17 0.20 0.17
[0.11] [0.10] [0.16]
Y(t- 1) -0.20** -0.20** -0.21**
[0.06] [0.06] [0.08]
Y{t- 2) -0.02-0.02 -0.05
[0.05] [0.05] [0.06]
SG(t) 0.20
[0.18]
SG(t - 1) 0.35*
[0.16]
G( t- 1) X SG(t- 1) 0.19
[0.12]
F-stat instruments 10.61 12.00 24.20
Observations 950 950 410
Notes: For each province i, the varia

that are in the same region as i, exclu
SGi t_i interacted with G, ,_h both mea
umn we show results where our origin
gate either two or three adjacent provi
region x year level and robust to hete
our p- values begin at 0.001, our schem
*** Significant at the 0.1 percent leve
spillover term in the set of re

of the coefficients on the con
umns 1 and 2 of Table 6).
The third column of the tabl
to small provinces with new o
The excluded instruments F-s
somewhat; that of Gi r_i is m
evidence on the spillovers effe
B. Influence of Individ
Some episodes in our sample

mates, similar to how certain
build up during World War II
multiplier effects. We address
evidence is sensitive to the ex
In our check, no single provi
In the first seven columns of
provinces in turn: Napoli, Cas
As shown in Table 1, these are
Table 7 - Dropping Provinces
NA CE PA CT SA BA RC
G(t) 1.86*** 1.47** 1.46** 1.35* 1.36** 1.53*** 1.37**

[0.44] [0.47] [0.46] [0.54] [0.47] [0.41] [0.43]
G(t - 1) 0.93*** 0.76*** 0.76*** 0.72** 0.72*** 0.78*** 0.73***
[0.22] [0.21] [0.20] [0.24] [0.21] [0.19] [0.18]
G(t - 2) 0.24 0.17 0.18 0.16 0.18 0.18 0.16
[0.12] [0.11] [0.11] [0.11] [0.10] [0.11] [0.10]
Y{t- 1) -0.20** -0.21** -0.20** -0.20** -0.19** -0.19** -0.17**
[0.07] [0.06] [0.06] [0.06] [0.06] [0.06] [0.06]
Y{t - 2) -0.00 -0.03 -0.02 -0.03 -0.02 -0.01 -0.04
[0.05] [0.05] [0.05] [0.05] [0.05] [0.05] [0.05]
F-stat instruments 19.59 9.48 11.31 10.90 9.25 11.85 9.42
Observations 940 940 940 940 940 940 940
Notes: Each column reports estimates after drop

CT: Catania; SA: Salerno; BA: Bari; RC: Reggio C
and robust to heteroskedasticity are reported in
for denoting significance is as follows:
* Significant at the 5 percent level.
dismissals. The point estimates of ß

level) are in the range 1 .35-1 .86, in a
on the first lag of public spending.
C. Influence of Heterogeneity
Time, and Province-Specific
In
Table 8, we show the effects of
only, and dropping the year and pro
ciple be consequential for our estima
estimates could be exposed to the i
cies, as well as aggregate cyclical fl
without province fixed effects, mul
effects discussed in Section V.
In the first column of Table 8 we show the effects of restricting the sample to
southern provinces only, as a way to detect whether heterogeneity across macro
areas could impinge on our results (see column "Drop north"). By excluding
Northern provinces, the coefficients of contemporaneous and lagged spending rise
somewhat (to 1.89 and 0.95, respectively). But so does, in absolute value, the coeffi-
cient of the lagged value of output growth (now equal to -0.34). Thus, our estimate
of the overall multiplier remains unaffected. In the second column we consider the
case of removing the year fixed effect (see column "Drop A/'). Also in this case the
coefficient of contemporaneous spending rises somewhat, to 1 .92. The point esti-
mates of the coefficients on lagged spending and output are, however, unaffected.
Similar results follow from dropping the province fixed effect (see the third column,
"Drop a, " of Table 8). In this case our point estimate of ß is 1.62. In conclusion,
Table 8 - Further Results
Drop north Drop Xt Drop

G(t) 1.89***
1.92*** 1.62***
[0.42] [0.52] [0.37]
G(t- 1) 0.95*** 0.75** 0.74***
[0.19] [0.27] [0.18]
G(t - 2)
0.23 0.12 0.12
[0.14] [0.13] [0.10]
y(i-l) -0.34*** -0.14* -0.11
[0.10] [0.06] [0.07]
Y{t - 2) -0.01 0.09 0.05
[0.08] [0.07] [0.06]
F-stat instruments 10.54 23.89 13.16
Observations 340 950 950
Notes: In the first column we rest

columns we drop, respectively, ye
the region x year level and robust
our p- values begin at 0.001, our s
*** Significant at the 0.1 percent
** Significant at the 1 percent lev
* Significant at the 5 percent level
none of these experiments

from those of our baseline estimation.
VI. Conclusions
In this article we have contributed evidence of a nonnegligible short-run output

effect of public spending at the local level. By relying on episodes of sharp con-
tractions in infrastructure expenditure in Italian provinces, we estimated the local
multiplier to be 1.5 on impact, and 1.9 including dynamic effects. We also find no
relevant spillovers of spending shocks in a province on the economic activity of
nearby provinces. By the features of our empirical model and data, these estimates
do not reflect budgetary and monetary policy interactions - interactions that argu-
ably play a key role in determining the aggregate output effects of deficit-financed
public spending at the national level.
The policy relevance of quantifying local multipliers is apparent. First, our esti-
mates shed light on the extent to (and the conditions under) which fiscal tools,
mainly through redistribution of resources, may provide effective instruments to
address area-specific downturns. Second, in times of crisis, financial and fiscal stress
may force local governments to implement deep, upfront cuts in spending, with
large variation in their extent across areas. Our estimates suggest that differences in
the intensity of the upfront retrenchment at local level can translate into significant
geographical variation in economic activity.
The Italian provinces we focus on in our study are akin to very small and very open
economies sharing a common currency. Our results may thus suggest that economies
with these characteristics may actually be quite "insular" in their dynamic response
to temporary variation in public spending. Analytical insights on the transmission
mechanism are provided by new-Keynesian models of

currency union or in a credible system of fixed excha
Corsetti, Kuester, and Muller (2013) - a point also
Steinsson (2014). These analyses put forward a mechan
models of small, open economies without monetary au
of the output multipliers around or above unity, for a
adjustment rules.32
Yet the analysis of local multipliers raises distinct issu
in open-economy studies. As discussed in Section VI of
cross-border spillovers at the local level may differ
"demand leakages" emphasized by the open-econom
requires a careful specification, at both the theoretica
spatial models accounting for cross-border mobility o
new promising direction for research could help in br
gate dimensions of the multiplier, by providing a fram
effects at national level of transferring resources acros
Data Appendix
Public spending: Our public spending variable is public investment, which

includes spending on the following categories: Transport (roads and airports, rail-
roads, and other kinds of transportation, ports and rivers, telecommunications);
Sanitation-Energy-Reclamation (hospitals, electric plants, swamps, land recla-
mation, other categories); Buildings (public buildings and schools; public spend-
ing devoted to private buildings). ISTAT provides a consistent data series on
current-prices spending on infrastructure at the provincial level from 1987 to 1999.
The source is: ISTAT, Annuario delle Opere Pubbliche (various issues). We deflate
the provincial public investment using the national GDP deflator for Italy.
Value added: The value added at the provincial level is measured in millions of
euros at current prices. We deflate the provincial value added using the national GDP
deflator for Italy. Source of value added: Istituto Guglielmo Tagliacarne. Source of
population: Italian Institute of Statistics, ISTAT (Statistiche Demografiche). Source
of deflator: ISTAT (Contabilità Nazionale). In particular, we construct the value
added growth rate by relying on two different series of provincial-level value added
for the periods 1985-1991 and 1991-1999, respectively. Data for the first period are
used to construct values of Y up to 1991, while for the period after 1991 we use the
more recent series.
Population: Source: ISTAT, Statistiche Demografiche (various issues).
32 In a currency area, an unexpected contraction in public demand in a region tends to reduce local prices in the
short run. Given nominal rates, this drives up the short-term real rates in the region. Because purchasing power par-
ity holds in the medium to long run, however, local prices are expected to rise back to the level prevailing outside
the region. Correspondingly, given nominal rates, future short-term real rates are expected to fall. These opposite
movements in current and anticipated future short-term real rates imply that the impact response of the long-term
real rates to a fiscal shock, arguably the relevant ones for private spending decisions, is actually quite small. As a
result, in the region hit by the shock, private demand is not crowded-in appreciably, and economic activity tends to
fall initially by the full extent of the unexpected fiscal contraction.
2208 THE AMERICAN ECONOMIC REVIEW JULY20I4
Employment: Sources: Istituto Guglielmo
Cassa Integrazione Guadagni: Hours of wag

Integrazione Guadagni," the main unemplo
employees of large private firms in Italy. S
Council dismissal related to the 1991 anti-

ties placed under the administration of ext
ernment on evidence of ties between admin
the direct infiltration of mobsters among lo
indirect influence. Source: Commissione
della criminalità organizzata mafiosa o sim
Council dismissal not related to the 1991 a

dismissals not related to Mafia infiltration.
ignation by elected officials; (ii) failure to
ineligibility of the mayor; (iv) failure to p
in the ruling coalitions; and (vi) other reaso
Mafia-type association: People reported by

ity because of mafia-type association (art. 4
ISTAT, Statistiche giudiziarie (various issue
Extortion: People reported by the police fo

extortion. Source: ISTAT, Statistiche giudiz
Murder: People reported by the police f

of murders related to the activity of Mafi
giudiziarie (various issues).
Corruption: Crimes and prosecutions for c

ment, misappropriation of public funds, ex
ISTAT, Statistiche giudiziarie (various issue
REFERENCES
Acconcia, Antonio, Giancarlo Corsetti, and Saverio Simonelli. 2011. "Mafia and Public Spend
Evidence on the Fiscal Multiplier from a Quasi-experiment." Centre for Studies in Economics
Finance (CSEF) Working Paper 281.
Acconcia, Antonio, Giancarlo Corsetti, and Saverio Simonelli. 2014. Mafia and Public Spend
Evidence on the Fiscal Multiplier from a Quasi-Experiment: Dataset." American Economic Revi
http://dx.doi.Org/10.1257/aer.104.7.2185.
Acconcia, Antonio, Giovanni Immordino, Salvatore Piccolo, and Patrick Key. forthcoming. Acco
plice* Witnesses and Organized Crime: Theory and Evidence from Italy." Scandinavian Journal of E
nomics.
Angrist, Joshua D., and Jorn-Steffen Pischke. 2009. Mostly Harmless Econometrics: An Empiricist
Companion. Princeton: Princeton University Press.
Barro, Robert J., and Charles J. Redlick. 201 1. "Macroeconomic Effects from Government Purcha
and Taxes." Quarterly Journal of Economics 126 (1): 51-102.
VOL. 104 NO. 7 ACCONCIA ETAL: MAFIA MULTIPUER 2209
Bertrand, Marianne, Esther Duflo, and Sendhil Mullainathan.

Differences-in-Differences Estimates?" Quarterly Journal of
Blanchard, Olivier, and Roberto Perotti. 2002. "An Empirical Cha
of Changes in Government Spending and Taxes on Output
117 (4): 1329-68.
Canova, Fabio, and Evi Pappa. 2011. "Fiscal Policy, Pricing F
tion." Economic Policy 26 (68): 555-98.
Cassese, Sabino. 1983. "Espansione e Controllo della Spesa Pubb
Politica Economica 73: 153-71.
Chodorow-Reich, Gabriel, Laura Feiveson, Zachary Liscow, and William Gui Woolston. 2012. "Does
State Fiscal Relief during Recessions Increase Employment? Evidence from the American Recov-
ery and Reinvestment Act." American Economic Journal: Economic Policy 4 (3): 1 18-45.
Christiano, Lawrence, Martin Eichenbaum, and Sergio Rebelo. 2011. "When Is the Government
Spending Multiplier Large?" Journal of Political Economy 119 (1): 78-121.
Clemens, Jeffrey, and Stephen Miran. 2012. "Fiscal Policy Multipliers on Subnational Government
Spending." American Economic Journal: Economic Policy 4 (2): 46-68.
Cohen, Lauren, Joshua Covai, and Christopher Malloy. 201 1. "Do Powerful Politicians Cause Corpo-
rate Downsizing?" Journal of Political Economy 1 19 (6): 1015-60.
Commissione Parlamentare di Inchiesta sul Fenomeno della Criminalità Organizzata Mafiosa o
Similare. 2005. "Documento di sintesi della discussione sulle problematiche concernenti la norma-
tiva sullo scioglimento dei consigli comunali e provinciali conseguente a fenomeni di infiltrazione e
di condizionamento di tipo mafioso." July 12.
Corsetti, Giancarlo, Keith Kuester, and Gemot J. Muller. 2013. "Floats, Pegs and the Transmission
of Fiscal Policy." In Fiscal Policy and Macroeconomic Performance. Vol. 17, edited by Luis Felipe
Céspedes and Jordi Gali, 235-81. Santiago, Chile: Central Bank of Chile.
Corsetti, Giancarlo, Andre Meier, and Gemot J. Muller. 2012a. "What Determines Government
Spending Multipliers?" Economic Policy (72): 521-58.
Corsetti, Giancarlo, Andre Meier, and Gemot J. Muller. 2012b. "Fiscal Stimulus with Spending Rever-
sals." Review of Economics and Statistics 94 (4): 878-95.
De Rita, Giuseppe. 1995. I Consigli Comunali Sciolti per Infiltrazioni Mafiose. Roma: Documenti
CNEL.
Dickie, John. 2004. Cosa Nostra. A History of Sicilian Mafia. London: Hodder & Stoughton.
Fishback, Price V., and Valentina Kachanovskaya. 2010. "In Search of the Multiplier for Federal
Spending in the States During the New Deal." National Bureau of Economic Research Working
Paper 16561.
Guiso, Luigi, Paola Sapienza, and Luigi Zingales. 2004. "Does Local Financial Development Matter?"
Quarterly Journal of Economics 1 19 (3): 929-69.
Ilzetzki, Ethan, Enrique G. Mendoza, and Carlos A. Vegh. 2013. "How Big (Small?) Are Fiscal Mul-
tipliers?" Journal of Monetary Economics 60 (2): 239-54.
Leeper, Eric M., Todd B. Walker, and Shu-Chun Susan Yang. 2009. "Fiscal Foresight and Information
Flows." National Bureau of Economic Research Working Paper 14630.
Ministro dell'Interno. 2000. "Relazione sull'attività svolta dalla gestione straordinaria dei comuni i cui
consigli comunali sono stati sciolti oer condizionamenti di tiDo mafioso" Seotember 1 9.
Misra, Kanishka and Paolo Surico. 2013. "Consumption, Income Changes and Heterogeneity: Evi-
dence from Two Fiscal Stimulus Programmes." CEPR Discussion Papers 9530.
Moretti, Enrico. 2010. "Local Multipliers." American Economic Review 100 (2): 373-77.
Nakamura, Emi, and Jon Steinsson. 2014. "Fiscal Stimulus in a Monetary Union: Evidence from US
Regions." American Economic Review 104 (3): 753-92.
Ramey, Valerie A. 201 1 . "Identifying Government Spending Shocks: It's All in the Timing." Quarterly
Journal of Economics 126 (1): 1-50.
Romer, Christina D., and David H. Romer. 2010. "The Macroeconomic Effects of Tax Changes: Esti-
mates Based on a New Measure of Fiscal Shocks." American Economic Review 100 (3): 763-801.
Serrato, Juan C. S., and Philippe Wingender. 2011. "Estimating Local Fiscal Multipliers." Unpub-
lished.
Shoag, Daniel. 2010. "The Impact of Government Spending Shocks: Evidence on the Multiplier from
State Pension Plan Returns." Unpublished.
Sims, Christopher A. 2010. "But Economics Is Not an Experimental Science." Journal of Economic
Perspectives 24 (2): 59-68.
Woodford, Michael. 2011. "Simple Analytics of the Government Expenditure Multiplier." American
Economic Journal: Macroeconomics 3(1): 1-35.

Acconcia, Corsetti and Simonelli - 2014 - Mafia and Public Spending - Evidence On The Fiscal Multiplier From A Quasi-Esperiment

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Acconcia, Corsetti and Simonelli - 2014 - Mafia and Public Spending - Evidence On The Fiscal Multiplier From A Quasi-Esperiment

Uploaded by

Copyright:

Available Formats

American Economic Association

Mafia and Public Spending:

By Antonio Acconcia, Giancarlo Corsetti, and Saverio Simonelli*

A law issued to combat political corruption and Mafia infiltration of

A key question concerns the efficacy of fisc

charges against, mobsters for Mafia-related crimes) ove

I. The Empirical Model

In our study, we aim to recover the short-run multiplicative effects of public

To carry out this study, we have assembled a d

(1) Y¡¡ = ßGu, + a¡ + A, + 7 X, , + v,-,,

where the coefficient ß measures the contemporaneous one-year government spend

the spending side, these laws gave the central governme

controversial, recent empiri

employment, and hours of wage supplement

II. Instrumenting Changes in Public S

Despite the advantages of our empirical m

8 The Cassa Integrazione Guadagni (CIG) is an Italian institut

A. The Institutional Setting: Mafia Infiltration and Co

We introduce our instrument by providing backgrou

Table 1 - Council Dismissals because of M ah a Infiltration

Napoli 48 Reggio C. 37 Palermo 23 Bari 5

Notes: The table reports the number of council dismissals because of M

differences across provinces, mostly driven by historic

B. An Instrument "One Can't Refuse"

When the government of a municipality is dismissed o

15 Statistics on convictions of the crime of Mafia association by regions an

Table 2 - Investment Spending in the First Year after Council Dismissal

Difference -19.65*** -0.46** -23.67*** -0.49* -4.72 -0.04

Notes: The table reports one-sided mean difference

treatment group amounts to about half

(iii) purchase of mechanical equipment; (i

C. Is the Instrument Variation Systematically

Police investigations leading to the dismissa

(2) Y¡ f = do + d i D¡¡ + dļt + dļ(t x D¡¡ ) +

Table 3 - Provincial Growth Rates Prior to Council Dismissals

Above average Below average Switching

t - 1 and t - 2 1/3 1/6 1/2

Notes: For any province that experienced cases of compulsory

equal to -0.07, with a standard error adjusted for c

A key feature of our instrument is that the time s

We implement our IV strategy accounting for two

20For provinces with repeated cases of dismissals, we consider only th

finance may be cut by the commissioners. Rel

(3) G¡ , = Õ ļ CDS 1 , i + ¿2 CDS2i t_' + o

III. Results: Impact and Dynamic Mult

The results from our baseline model are shown

Table 4 - Public Spending Multplier

OLS 2SLS 2SLS

First Second First Second

G(t) 0.21** 0.23** 1.46** 1.55***

Observations 950 950 950 950 950 950

Notes: Data are annual from 1990 to 1999 at the

estimate of 1.46 or 1.55, depending on w

Wingender (2011) and Nakamura and Stein

24 As pointed out by one of our referees, potential differen

IV. Do City Council Dismissals Affect Output

For the proposed IV estimation to be reliable, our instr

A. Variation in Mafia Activities

The exclusion restriction might not hold if the empir

26Country-level studies suggest a negative relationship between the diffusi

the short run. On the other hand, the Mafia ma

(4) Yit = <5, CDS'it + ô2CDS2it_x + a¡ + A, + çXu +

Difference -19.65* -0.46 -23.67*** -0.49* -4.72 -0.04

G(t) 0.21 0.23 1.46 1.55*

G{t) 1.30 1.56* 1.53* 1.59* 1.64***

G(t) 1.86* 1.47 1.46** 1.35* 1.36 1.53* 1.37**